Drill to Detail - Drill to Detail Ep.100 Special ‘Past, Present and Future of the Modern Data Stack’ with Special Guests Keenan Rice, Stewart Bryson and Jake Stein

Episode Date: March 7, 2023

Joining Mark Rittman for this 100th Episode Special are Keenan Rice (previously Founding Team, GTM Executive at Looker, now GM at Firebolt), Stewart Bryson (previously CEO at Red Pill Analytics, now Head of Customer Experience at Coalesce) and joining us mid-way through the show as our even more special mystery guest, Jake Stein (previously Co-Founder and COO at RJ Metrics, CEO and Co-Founder at Stitch and now CEO and Co-Founder at Common Paper.Drill to Detail Past Episodes discussed on the show were:Drill To Detail Ep.1. 'After The Gartner BI&A Magic Quadrant 2016', With Special Guest Stewart BrysonDrill to Detail Ep.23 ‘Looker, BigQuery and Analytics on Big Data’ With Special Guest Daniel MintzDrill to Detail Ep.33 'Building Out Analytics Functions in Startups' With Special Guest Tristan HandyDrill to Detail Ep.71 'The Rise of Snowflake Data Warehouse' With Special Guest Kent GrazianoDrill to Detail Ep.69 'Looker, Tableau and Consolidation in the BI Industry' featuring Special Guests Tristan Handy and Stewart BrysonDrill to Detail Ep.89 'Firebolt and the History of Cloud Data Warehousing' with Special Guest Eldad FarkashDrill to Detail Ep.99 “Is the Modern Data Stack Dead?” with Special Guest Chris TabbOther show notes below:Gartner Makes it Official: The Age of Self-Service Is Upon UsRJMetrics Acquired by Magento Commerce, Pipeline is now StitchThe Startup Founders’ Guide to Analytics - Tristan HandyAnalyzing Drill to Detail Podcast Stats using HexGoogle Completes Looker AcquisitionCommon PaperRittman Analytics : How We WorkDrill to Detail Ep.96 'Omni's Mission to Answer the First 100 Questions' with Special Guest Colin ZimaCoalesce : Start Free

Hey, it's great to be back, Mark. Really appreciate that you're recording again. So that's exciting. I saw that in my podcast and was really excited that you're back. I'm excited to be on the 100th episode. I think I might be the most interviewed guest on this podcast, but I'd have to go back and look. And really excited that Kenan's here joining. You know, we go way way back and keenan it's really great to be on this episode with you yeah thanks for uh finally inviting me after 100 episodes mark uh yeah well we ran out of interesting people to interview just kidding so we scraped the bottom of the barrel and we found you keenan exactly the last the last data guest to invite to the podcast was keenan rice well anyways regardless i like to be in the barrel so thanks for having me on the podcast today looking forward to it So welcome to this special 100th episode of the Georgia Detail Podcast, and I'm your host, Mark Rittman. So we're joined today by not one, but two special guests, Stuart Bryson and Keenum Rice. Keenum, great to have you on the show. Why don't you tell everybody what you do?

Yeah, what is it I do is a question a lot of people have asked me over the last 12 years. Yeah. So, I mean, probably most famously helped start Looker with Lloyd and Ben. It was the founding business guy over there. Title was business guy in 2012. So I helped sales, marketing, all that fun stuff. I know all you guys because I also built and ran our global partner ecosystem. So thanks to the work that all of you guys did,

building amazing advanced boutique consulting companies really helped Looker and I think the whole modern data stack start getting adopted and taken seriously and shown what the power of it could really do. Yeah. And then after that, I've just been doing a ton of tech investing, obviously a lot on the data side and across the rest of enterprise software. And most recently also been helping the guys at Firebolt expand to the US and keep growing their business as well. And Stuart, so as you said, you're the guest who's been on the show most times out of anybody. But if anyone doesn't know you, maybe just tell us who you are. And again, what do you do? Yeah, so I think every previous episode, I was at Red Pill Analytics. And that's a consulting company that I founded with a good friend, Kevin McGinley, who went on to Snowflake probably at the exact right time.

And I sold that in 2019. But during my time there, we initially started as an on-prem Oracle consulting company. As soon as we learned about Snowflake and Looker, we pivoted. We had a lot of customers sort of in the barrel that were looking to move to the cloud. So, you know, honestly, we were just a on-prem to cloud migration company for seven years. And, you know, worked closely with Kenan and the folks at Looker. Shout out to Rich Christie, who was also on the partner side. Amazing. side. It was amazing. And I really appreciate that Keenan and Rich saw in us at Red Pill the ability to execute on their vision and really helped us grow. They were the first modern data stack company that really believed in us. Since selling the company, I did my earn

out with them, bounced around a little bit, helped start a data quality startup called Qualytics. Shout out to those guys. Recently left so that I could move to Coalesce. And Coalesce is a data transformation tool, a nice injection into the modern data stack. I run all of customer experience there. So that means support rolls up to me, customer success rolls up to me, and DevRel rolls up to me. And I've been with them since September.

I really love the product. I love the opportunity to, you know, bring a little bit of those, you know, what we saw in legacy tools. There was a lot of crap in legacy tools, but there were some things worth keeping. So this is a product built from the ground up to support the modern data stack, but also bring a little bit of flavor from the enterprise that I think we're missing in the modern data

stack. So as I said, this is actually a special episode, our 100th episode. And as well as our two main guests, Stuart and Keenan, we've actually got a mystery third guest coming on in about 20 minutes as well as our two main guests uh stewart and keenan we've actually got a mystery third guest coming on in about 20 minutes as well so um so what i'll be trying to do is obviously over the next 20 minutes trying to get the two of them to say uh terrible things about our third guest and then i can invite them on and uh it'll be nice and awkward for the rest of the show just slack just slack me mark and tell me who you want me to demean i'll get right on it thanks stewart the theme for this special 100th episode is the past, present and future of

the modern data stack and we use as our starting points for discussions some past episodes of the show that were looking back actually particularly significant and particularly kind of historically interesting and that would be our starting point for each conversation. So I'll add links to each of the episodes we talk about in the show notes on our website which is ripmananalytics.com forward slash drill to detail or you'll find them on the iTunes show notes and episode notes and so on. So let's start then by going back to back in fact to 2016 the very first episode of the show where Stuart joined us for our first ever guest and he was talking about what was the hot topic in the world at the time which was the recently published Gartner BI industry report that had

just excluded Oracle, that was the main tool we were using at the time, from that report's magic quadrant and was saying that the future is no longer about semantic models and governed BI and so on. It was actually more about desktop BI discovery and what they were calling bimodal, you know, bimodal IT and bimodal analysis. So Stuart, tell us a bit more about that time and what was in that report. That's the first time I've heard bimodal, probably in five years. So if that is any indicator of how predictive that report was. But basically that report talked about enterprise analytics tools or BI tools were dead. And the only thing that was going to survive was the more user-focused, we'll say worksheet-type products, those that could be used outside of IT. And it's just not what happened.

Kenan's going to be able to talk about the success they had at Looker. That's certainly not what Looker was about. And it's why we loved Looker. It looked like some of the tools we were used to, but with a modern framework and architecture. And if you look at things like Tableau and Power BI, they've moved more and more toward enterprise features because that's's at the end

of the day what people want to deploy for these huge data platforms, data clouds, data warehouses, data lakes, whatever you want to call them. So that report was basically firing a shot across the bow for all these enterprise tools and it simply didn't happen. So is that enough context there, Mark? Yeah, that's great. Thanks. So getting on to the first of the historical episodes I want to talk about, what we're going to do is go back to episode 23, Looker BigQuery and Analytics on Big Data with special guest Daniel Mintz. And so this was the first ever episode that we talked about Looker on the show. And the context for this was that I was working

at the time at a startup in London called Qubit, where we used Looker on top of BigQuery to serve up personalization data and kind of web analytic data at huge scale. When I first started working with Looker, it took me a little while to get my head around some of the terminology like LookML and Explores and so on. But then it gradually dawned on me that this was another, it was just like a modernization, a reimplementation of some of the concepts that I was very familiar with from the Oracle BI world, so semantic models and so on. So maybe start off with Keenan. What were your thoughts at the time when Looker was being built and about, I suppose,

semantic models and metadata layers and so on? And what were your thoughts about the Gartner report at the time, the one that talks about bimodal IT and desktop BI taking over? Yeah. So I think Stuart did a really good summary there of that report. And so that was an interesting time, right? Obviously, I get why they wrote that report, right? Tableau was just insanely growing, right? And everyone wanted this really easy to

use, really beautiful way to analyze a lot of their disparate data sets, right? That were all around the organization and they didn't really get access to this. They meaning the business users from a central data team, right? So I actually like that report in the sense that like it was, it over indexed so hard on that, that it made the data teams really take a look and be like,

I don't know if this is totally true, but from my perspective at Looker, like, holy crap, like we got to get our stuff together, right? How do we actually become more effective within the organization? Because data is hard, right? And data teams are there for a reason, right? And so I think it was a really cool, serendipitous kind of time in which a whole bunch of interesting technologies like Looker, like Snowflake, you know, early on with Redshift and what Honorog did at AWS Big Data

in general to put all this stuff together and say, okay, now there's a whole bunch of tools. Here you go, data teams. Step up now and be the service organization you should be. So I think over-indexing on that in the Gartner report and just in the industry in general probably actually was really good. And it was probably maybe one of the best things that happened and helped be the impetus for the modern data stack in general. So while I think that was exact opposite, and it was very hard

for us to get any time of day with Gartner at Looker, because we were the exact opposite trend they wanted to see. I think it all worked out very well in the end, mainly for, you know, folks in the organization that wanted to access and analyze data. Stuart, what was your kind of reaction to Looker when you first came across it? Well, I'm going to be completely honest, Keenan. You know, there were features that were missing that I thought I needed. That was the first thing I noticed. Federation, bringing in multiple databases and aggregate awareness and blending and all those things that were so important when we were constrained by technology on-prem. So my initial thought was, I love the Git integration. Mark, I think you famously tweeted something around

Stuart Bryson needs to see Looker or something along those lines, because you know how much I loved Git back in the day. But then I realized as we started using it, that actually those features are not as important in cloud data warehouses as they used to be. And I think that that's one of the things that the Gartner report, kind of that over-indexing is that they weren't really also cognizant at the same time that with federation, not federation, but with the consolidation of database of data into a single database platform, we didn't really need those features. So it was refreshing when I finally understood that and started using Looker the way it was intended to be used. We delivered projects way faster. We delivered better projects.

We weren't dealing with plumbing like we used to back in the day on-prem. We were just able to do things with data. That's funny to say, but I got into this industry to do interesting things with data. I mean, that's funny to say, but, you know, I got into this industry to, you know, do interesting things with data, but on-prem I was installing and integrating and doing plumbing is the best term I can use. So what I really learned as we did more and more projects with Looker is just how it really covered 90% of what we needed. And all those features that I thought I wanted, I didn't need, but still kept the one very important one, which was the semantic layer. And I still think, and I still believe in that semantic layer. That's one of the things that

I brought from my legacy world that I didn't want to give up. And, you know, Looker really, really covered that well. Well, even today, it's just proven to be the number one thing you did not want to give up, right? And should not have ever been given up, for sure, right? Is that semantic layer. So what was it like in those early days of Looker? Was the tool being built for maybe an enterprise audience or was it being built for a different type of audience maybe?

Maybe one that was more, I suppose, technical or more kind of digital first and so on? Yeah, the initial first years were really fun. You know, Lloyd Tabb, who founded Looker and brought on Ben Porterfield, who was the VP of engineering. They're engineers, right? And so in 2012, you know, they built a tool for themselves to analyze data, right? They didn't come from the data world, right? They built an engineering tool for engineers that needed to deal with data. And I think that was, you know, just fortuitously, like 2020 foresight, that so many trends were going to kind of be really large secular trends around that core concept that I wouldn't say we invented by any

means, but I would say that we were very well positioned to do that, right? So for instance, Lloyd is not just a software engineer, but also a languages engineer, right? And so he just built this tool to say, hey, I want to describe my data. And that's living in a database. And then I'll, you know, we'll build this modern app that Ben came in and helped develop to say, you know, we'll give you a web interface to basically, you know, ask and answer questions of your data in a way that you can describe, you know, more powerful or more complex concepts that exist in the physical schema. And it was like, oh, you've built a BI tool with a semantic layer, but that's not how

they started it. So it was really cool because a few things that come out of that is obviously, we've just discussed the semantic layer in detail. But the way that he approached it in an engineering language really became the right paradigm to analyze data that would then be stored in a completely different way that was not the way it was stored really kind of in 2012 when we started, right? But that consolidation, you know, kind of original fidelity kind of data, very large wide tables, this kind of, you know, kind of the consolidation of how people use cloud data warehouses today, right? That paradigm of the semantic layer that Lloyd had written was perfect for that. And then the second kind of architectural component of just

analyzing the data in the database, we kind of reverse what was the way of thinking with Oracle BI, right? Of pulling it out, bringing it into cubes, right, and then analyzing those smaller sets of data, really paralleled well with the fact that cloud data warehouses became the way to analyze data, right, because it was faster, it was cheaper. And then there became a whole ecosystem of tooling to put all the data inside of that, right. So when the technology met up with that paradigm of the data model, it really, kind of a whole bunch of really great secular trends and great foresight on Lloyd thing in the world to just tell someone, just trust me, it's better. Like, that's like the hardest thing to say. Yeah. Because, you know, it's like...

Perfect timing, wasn't it, Keenan? I mean, it was really perfect timing when the cloud data warehouse caught up, right? Almost at the exact right time. It was like literally perfect. So we built Looker around in the era of either a replica of RDS MySQL or Green Plum Vertica and Park Cell, right? And so we're just scouring the Vertica group. So like Carl Mahoney and Chris Selland and stuff like that

at every Vertica big data show, like just trying to get all these new customers. And we were introduced to Anurag Gupta in 2013. And he's like, I got something that you guys are going to really love. And it was Redshift, right? And the second we just saw an MPP database in the cloud, self-managed, we're like, oh, I remember Lloyd just being like, this is game changer like this is it we're all in and then when we met the snowflake team because we were both redpoint series a companies um we were like oh holy holy shit this is this is even better um yeah so that was that was all really cool for fortuitousness in like 2014.

Now the mystery guest we're going to bring on, just like to reassure you, isn't Thomas Kurian. But just to ask, how much of this success of Looker was down to Looker the company and the culture of the company? Yeah, I think probably just as much as the amazingly large secular trends that we were able to ride. The culture that was created, I think, was bar none, you know, one of the coolest things I've seen. And I've invested in well over 100 companies. And I think some folks have been building amazing companies as well. And just being part of Looker

for 10 years was very cool. So I think the culture internally, but also the culture we had with our customers. So Stuart, you were talking a lot about that early on with just making customers successful and, and being super happy about that. That was core, right? It was very, very, very core. Everyone did anything and everything at Looker to make customers successful and stuff like that. And partners. And partners. Yep. That's true. We had a very large partner team as well. So we just wanted everyone to know that they could have a really, really, really great experience. And we knew we could control that. And so we invested a lot in there across the board. Moving on then to the next episode, we want to talk about the history of the modern data stack

and drill to detail. It was actually episode 33, which is building our analytics functions in startups with special guest Tristan Handy. So this was Tristan before dbt Labs, and it was actually Fishton Analytics. We were talking about analytics maturity levels, but also talking about DBT. And so this is actually quite a

good opportunity now to introduce our mystery guest, who isn't Tristan, but maybe I'll let you introduce yourself. Yeah, many people think of me as a poor man's Tristan Handy. Very proud of that. Hey, this is Jake Stein, formerly of Twitch and RJ Metrics. Good to see y'all, or talk to y'all. in handy uh very proud of that uh hey this is uh jake's guy importantly of uh yeah and our geometrics good to see y'all or talk to y'all jake i recognized that voice the second you introduced yourself and it made me very happy i did i was hoping you would that that joke was talking today that's really just and stuart you know jake as well don't you absolutely i know jake back from the stitch days and also uh we're both investors in qualitics and so spent some time there

and uh yeah it's really that's why i knew his voice so well so jake great show from the show um and for anyone doesn't know you maybe just tell the story of stitch and uh remind me again how how um rg metrics led to all of Metrics led to all of these kind of companies. RJ Metrics sold to Magento, which then got acquired by Adobe, and Stitch sold to Talent. Yeah. So at RJ Metrics, we had the challenge of competing against what Keenan was just describing, which was challenging sure uh and i mean with tons of respect for for the liquor team in general and they made some like really fantastic early bets like he was saying like i remember i looked at the um the early deck that first round published or maybe

it was like an email of lloyd originally pitching it one of the bullet points and this was like i don't know 2011 2012, it was columnar databases are the shit. And the fact that that was one of his headline things back then is just an amazing amount of foresight, which candidly, I was nowhere near that early on. So yeah, RJ Metrics was an integrated BI product. Really, before the model of the modern data stack was created, we did data collection, data transformation. We managed to do a warehouse for our customers and we had our own proprietary biz layer and it was focused primarily on e-commerce. And essentially what happened is we were pretty early to the world of data,

at least for startups, we got started in 2008. And essentially what happened was we got a ton of initial traction. Lots of people used it and loved it. And then over time, years later, we eventually started to see some of our smartest customers start to churn in a very particular way. And that was very interesting to us where they would, they were going to BI tools like Looker. And they would say to us, hey, we're going to use them now, but any chance we could use that backend that you guys built. And I spent a lot of time trying to convince them that that was not what they wanted. And a lot of that was unsuccessful. And eventually we sort of took yes for an answer and thought, oh man, maybe this fact

that we are collecting this data for them is something we should like have as a separate product or at least allow our customers to access. So we went through sort of an iterative process of first, just like opening up access to our warehouse, which originally was on MySQL and eventually got migrated to Redshift. And then sending data into our customer's data warehouse. But it was always coupled to the BI product. And there was a big tension in the business model, which was that these two products represented very different philosophies on how analytics should be done, the closely coupled versus totally separated. And around the same time, Magento, which was our biggest partner, that was the most popular

platform that our customers were using to power their e-commerce stores. They got serious about solving their analytics and reporting challenges internally rather than through partnering. So they started a conversation with us around acquiring the company and there was an obvious strategic fit with them with the original RJ Metrics product. But this small second thing that was a separate product that would eventually become Stitch that would help you get data into your data warehouse was like 1% of our revenue. And it was very clear that was not a strategic fit for them. They just didn't really care about it. And so we were pretty sure it was going to die inside of Magento if it got acquired. So one of the things that we basically were able to do in that deal was to say, okay, the original business with 99% of the revenue and 80% of the team, you acquire that,

and then let us keep this other product and we spin it out into a separate business. And that's ultimately what we did and that became Stitch. And then the other thing that was sort of happening in parallel in the little skunk works within the company was Tristan and Drew and Connor and also Chris Merrick, who's our CTO, who's now at Omni. They were all working on this different way of working with the data that this proto-stitch product was provisioning around doing in-database transformations. And I remember one time Tristan came to a conversation with me and Bob, and Bob was my co-founder. He was the CEO of RJ Metrics and basically said he was thinking about

leaving and working on analytics and this DBT thing full time. And we said, that sounds pretty cool. You know, like we were supportive. I had no idea it was going to become what it became, but it was all out of just like doing analytics on some internal projects and thinking about a

better way to to solve the problem and really following that that sort of philosophy of you know analytics is a type of software engineering and ought to use similar tools and the open source model and things like that i'll pause there that was a lot i I want to just add in one thing real quick, Mark, is kudos to Bob and Jake. I think it is a beautiful case study on corporate, on leadership really, right? And strategic thinking about what the entire story Jake just said when him and Bob reached out to us and told us about Stitch, right? It was just incredible like you know the really break all that stuff apart be really authentic with who they are what they built and everything like that so we just had the utmost respect for for that decision so you know big big thumbs up and i

think it was obviously the right decision so keenan how much was the success of dbt really down to the lack of incremental refresh for PDTs in Looker? I think I still have nightmares about customers and partners like yourself requesting incremental PDTs, and I'm just stuck in this void of not being able to answer you. Yeah, thanks for that. No, I don't think it's only successes based on incremental PDTs. But what it did show, right, is it did show that there was a really, you know, a big opportunity to extend the modern semantic data modeling layer and extend meaning, if you think about the spectrum of data transformation in the modern world, you know, not everything had to live logically, right? And I think dbt really challenged that in a really positive way for the entire industry, you know, where PDTs were basically doing that stuff. I mean, they were materializing it, right? But you were doing it kind of like, on the other side of the BI tool or inside of the BI tool,

not prior to the BI tool, right? And I think the amazing thing, the amazing innovation that dbt brought was like, hey, there's a paradigm that I do think LookML definitely innovated and invented in that sense to pull this out and say, you can have a logical third-party extension here to really help you be more successful with data, especially large sets of untransformed data before they hit any tool.

And so I think that the creation of dbt actually, in my eyes, strengthened a lot of the power of LookML because it did take those things out, you could prototype now, you know, really interesting table level transformations and things like that with Looker. And you could use PDTs for smaller things. But you when you really wanted to productionize it and scale it and things like that with Looker. And you could use PDTs for smaller things, but when you really wanted to productionize it and scale it and things like that, dbt became a really natural extension to the ecosystem, right? And so that's why we were always really, really behind dbt and what Tristan was doing even at Fishtown with his consulting. So I'll let Stuart speak now. Stuart, when dbt was around, I suppose the other part to all of this was the move towards what's called analytics engineering.

Okay. So maybe just tell us a bit about what that is and what your thoughts were on that when it came into sort of like vogue. The first thing I want to say is that I'm shocked that Jake accidentally created the connector market. I mean, to hear that you initially thought that was a bad idea and then you helped, I'll say, create the connector market, which is just crazy because it's one of the biggest markets right now. And it enabled all of this, right? I mean, everything downstream from it, it was hard because it was tough to get data into cloud data warehouses. So first shout out, Jake, for you accidentally, you know, stumbling upon a huge market. But thank you. So just imagine what I'm missing right now. It'd be so exciting. But secondly, yeah, I mean, when we saw DBT for the first time, it really clicked with me because I had been trying for years to, you know, implement software development principles in tools that simply didn't support it.

And this is why I loved Looker and DBT the first time I saw them was they both, you know, one hunt. Maybe it's because they were built by engineers and not data people. I think there's give and take there, right? There's some things that seemed obvious to me that should have been there in both products. But at the end of the day, the, you know, doubling down or double clicking on the software development experience is something that I've tried to do with legacy tools for years. So it really clicked with me. And I think the idea that putting that sort of rigor testing, Git integration, CICD and data ops and all that, that I had wanted in the Oracle world and never had. It was just fantastic. Now, I had a whole team of people at Red Pill Analytics that were hired for on-prem work and now had to suddenly be converted

to the modern data stack. That was a challenge, but it was a fun challenge trying to get a bunch of consultants that weren't used to these life cycles, we'll say software development life cycles, instructed and implemented in how to do that. That was probably the thing I'm most proud of at that organization is being able to not only take consultants and employees, but also customers into that journey and understand the value there. So if there's one thing the modern data stack does that the legacy data stacks didn't do is that whole software development lifecycle, something that I've been pushing for for years. So that's the first thing I noticed about dbt and looker that immediately had me hooked

was that. Yeah, and I want to jump on that too, a little bit too. I know we talk a lot about BI too, and like modern BI and how this, you know, this kind of opened that up. But, you know, the big thing too, and this is why we also love dbt and just love the whole ecosystem growing is we didn't, we weren't just protective of lookers, the BI tool, right? A big part of, of our whole thing, as well as the data experience is going to live outside of the BI tool a lot of times too. And it's also why the consulting ecosystem was so important for us is help, you know, modern data teams figure out how they really need to use data within the organization, not just in the constraints of a BI tool UI, right? So when you get this software development, this modern software development approach into all the tools in the ecosystem,

then these teams, these data teams can start thinking like software engineers and like product managers. And they can start thinking about all the different things. Is there workflows that they can build in near real time of data going from the Stitch kind of connector world in through DBT, in through LookML and being piped out via an API and doing some sort of recursive process or something? Or is there some application that we could develop that's going to help internal teams work with suppliers

or provide data to customers or something in a much more effective and lower latent and more interactive way. So all of these outputs came on the backside of that as well, which I think, to me, probably really solidifies the value of the modern data stack. Because if it was just a better BI tool, it's cool. It's great. It just feels incremental to me. But the real value I've seen in a lot of these large customers is really making data more effective with these modern tools in this modern paradigm. Going on now then to the third episode we're going to talk about in this tree here is it's going to be episode 71, the rise of Snowflake Data Warehouse

with special guest Kent Graziano. I think this is actually probably the most popular episode we ever had. So Stuart, you've had a lot to do with Snowflake over the years. It's had quite a big impact on your business and your career. Just tell us a bit about what you think about Snowflake and why it was such a success in the market, really. Absolutely. I can tell you that Kent was trying to decide between joining Red Pill Analytics and this scrappy little startup called Snowflake at the time. And he chose wisely, let's be honest. Big win. Huge win. You know, I think I was close to convincing him.

But at the end of the day, I actually said he should go to Snowflake because even then I knew it was, you know, if the promise was fulfilled, it would be game changing. And I remember, you know, Kent is the reason that Red Pill got converted basically from an on-prem to a cloud consultancy was, you know, he enabled me to connect with a lot of the people at Snowflake. I remember giving a presentation to the entire company about integrating with Oracle tools and doing Oracle takeouts to some degree. And I don't know, it was 50 people, 60 people. I can't remember. And my math may be way off there. So once we did a project, we did our first project on Snowflake. It was amazing how quickly we turned things around.

We spent so much time on-prem installing and configuring and integrating, and all of that went away. It drastically changed the experience our customers had. I got on a soapbox after that project. And even customers who didn't want to hear about it, heard about it, because I wanted to tell them about the successes we had. And we were actually doing things with data. We weren't installing platforms that could maybe eventually do things with data. We were doing things with data. And so, yeah, it made me honestly turn my back on the Oracle world. I'm still in their highest tier of advocacy. I think that probably ends today. But I completely turned my back on on-prem. So I can't believe that you're still an Oracle director. I had to give mine up years ago, but you're still one even director. I had to give mine up years

ago, but you've still, you're still one even today. We knew he was going to make someone mad this episode. Here he goes. It's the entire organization now. I honestly have no idea why I'm still in there and I'm sure I'm not going to be tomorrow, but yeah, so that it drastically changed. And you know, the whole world's caught up. It's, it's, it's not like, you know, this is, I unlocked some mystery. But, you know, shout out to Kent for going to Snowflake instead of Red Pill. And then, you know, being a really good friend and immediately introducing us and getting us to drink the Kool-Aid because we absolutely did. It was the right thing to do. And Looker came maybe a year later and we totally drank that Kool-Aid. And, you know, DBT maybe a year later or two years later,

and we drank that Kool-Aid, and it worked out for our customers. Jake, how much did the success of Snowflake's IPO and the fantastic valuation that Snowflake got really in the market. How much did that influence your decision to sell Stitch to Talent? Sure, yeah. And so just going back to Stuart's comment on stuff that I've done by accident, one of the things which I also just did not appreciate at the time was what Snowflake would become or the level, how much better it was than the other options that were out there at the time was like what Snowflake would become or that the level, like how much better it was

than the other options that were out there at the time. Because I remember talking to Walter O'Donoghue in the very early days of Snowflake and they had a real challenge around getting data in to Snowflake. And this is something like speaking of people who've helped create the connector market, obviously like Taylor and George at Fivetran deserve a ton of credit for creating that category as well. And one of the things they got right before we did was that they saw the potential of Snowflake and became an early partner there. And that, you know, I have a lot of respect for getting that call right early. But it became obvious pretty soon that this was like a big force that was growing really quickly. And like, again, like following what a lot of our smartest customers were doing when they were, you know, let's say hitting the limits of Redshift or just, you know, trying to scale up in general. And then, yeah, even more so with snowflake's um ipo and subsequent valuation but like we had inklings that you know of the scale before that just through being a partner and

knowing some of the folks there um and like in some ways snowflake really drove what was at least the catalyst for um you know like the stitch, I think it was like Talon's efforts to partner with snowflake. Um, and some of the challenges they had there just because they were built for a very different model. I think when Talon did reach out to us, that was at least partially inspired by the fact that they saw that that was going to be a really critical partner to have, and that they could not, you know, get all the way there without, without someone like us. So, uh, yeah, very, uh, very impactful for, for a lot of reasons. Yeah.

Well, I mean, just extending on that, uh, let's talk, talk about impactful. I mean, looker, looker wouldn't even work if we didn't have the connector ecosystem, right? Because that was the big bet that we took architecturally is we weren't going to get around that and build our own connectors uh and i talked with lloyd and ben massively early on like come on why don't we have a sales force connector why don't we have a sale like just give me something to sell against rj metrics this is dying out here yeah that was a good one yeah thankfully uh taylor and george came around and i think unfortunately think, unfortunately, for the early RJ,

that was probably the biggest thing that unlocked the value of Looker, right? And then when Stitch also came out, then it kind of really solidified the fact that, you know, we can have this best-of-breed tooling across the data stack, right, and allow people to have choice and allow them to start piecing together stuff because we did, right? We just talked about the analytics engineer and this was the big secular trend there was modern data people wanted to be more like engineers than they did like spreadsheet people, right? And so they really wanted to do code and all of that.

And so they want to go pick their tooling uh and go build their stack together right and i think stewart that's why you and even mark too with your consultancy all did amazing jobs because you you were there along the way to help them start piecing this thing together and they had so much choice um so i think it was the ecosystem that made everything really successful here um and you know stuff like you know massively benefited from it which is amazing back to stewart again um stewart why do you think snowflake was was actually so successful i mean there's been database there's been database startups before snowflake there's been since snowflake um in some respects snowflake was was actually kind of one of the least ambitious um i

suppose new databases out there in that it was very similar in the way it worked to old style databases like Oracle. Quite different to say, sort of BigQuery, which had quite a different sort of model really for the way it worked and the way you administered it. So why did Snowflake do so well, do you think? So a couple of things. One is business organization and the other is technology. So I'll start with technology. Number one, it really is load your data and query. I mean, there's, you know, zero to snowflake is a thing at almost every consultancy, right? There is still some foundation delay usually around security.

But at the end of the day, you didn't have to worry about most things. And BigQuery just wasn't that way. You had to design tables in BigQuery. Also, I remember, here's another shot across the bow, I guess. I remember meeting with the office of the CTO at Google early on. And it was a guy from, I can't remember his name. It was a guy that had come from Spotify, I believe. But when you went into those sorts of roles at Snowflake, it was people from the enterprise, right? Yes, they had really great non-data background engineers building product, but they also layered in a lot of expertise from the data world. And I think that Google does things the Google

way and BigQuery shows that. And I think Snowflake said, hey, there's a basic engine swap that we need to do here. And it's separation of compute and storage. But a lot of what data teams expect, especially those coming from on-prem and enterprises, expect a lot of the stuff that you see in Snowflake. And it's stuff that looks a lot like Oracle and Teradata. Whether you could migrate, if you didn't have to worry about migrating the enterprise, yeah, sure. Maybe those features don't make a lot of sense. But if you want to migrate the enterprise, it's got to look and feel like something that they expect. And at that time, BigQuery simply didn't. At that time, you couldn't even do DDL

in BigQuery, right? It was an API call. So now obviously BigQuery's caught up. Their roadmap looks a lot like Snowflake's roadmap. They're just six months behind or a year behind. But I don't necessarily think that's true now but snowflake knew exactly what it would take to get the enterprise to migrate and let's face it at the end of the day that's where the dollars are spent okay so moving on to the next episode we can talk about which was around the same sort of time actually and it was episode 69 look at tableau and consolidation in the bi industry and that had Stuart on there and it also had Tristan on there as well and obviously myself and it was around that time when the kind of wave of acquisitions

was happening really within the industry where we had tableau being bought by Salesforce and of course more famously looker byer by Google Cloud Platform that had been announced at that time. So Keenan, you'd obviously been pretty central to a lot of what was going on when the acquisition happened. So what was it like really being at the eye of the storm really when that all was happening? Yeah, yeah, yeah. Definitely was part of the creator of the eye of the storm there. So yeah, look, I mean, it was a really interesting time obviously uh you know snowflake was thinking about their ipo we were actually getting ready for ours um so we were all definitely at scale pretty large sized um and you know we weren't looking to sell the company

or anything but uh but it was very clear so google approached us and when we started talking and discussing internally and externally like you know it it was very clear. So Google approached us and when we started talking and discussing internally and externally, like, you know, it was very clear that like, there could be some really amazing synergies with a BI tool being part of a cloud, you know, for all that massive distribution and hyper distribution scale. And also like a lot of stuff that we really wanted to do with Looker and keep expanding on Looker. So, you know, when we did the acquisition,

the core strategy was, you know, 100% stay open, stay across multi-cloud, be the thing that, you know, is still Looker, but grows with the distribution. So, yeah, it was a super exciting time for sure because, you know, we wanted to get Looker, you know, in the hands of a lot more people and doing so under Google was ultimately that decision for sure. Back over to Jake then. Jake, what was your reaction at the time to

the announcement that Looker was being acquired by Google Cloud Platform? And again, how much did that push you towards thinking about selling Stitch to Talend or just generally sort of like how the market was consolidating? So to answer the last bit first, it really didn't impact our thoughts about getting acquired or staying independent. I thought it was obviously extremely interesting and I was jazzed for all the people at Looker since it was such a great outcome. But from our perspective, it didn't change our competitive dynamic per se or our view of the market. I think it was... It always made a lot of sense to me.

And this is one of the great things about the Looker Google or Looker Snowflake partnership. And this is definitely a credit to Keenan and Norris and folks like that who worked on that team, was just that they drove a ton of revenue for the partner because they were selling compute and they drove a lot of compute.

And that's also, I think, to a degree why DBT, in addition to just their wide adoption, is such a valued partner by data warehouses today because they drive a bunch of compute um so it was like it may it was uh like shocking but unsurprising if that makes sense like i i didn't see it coming but it was like like instantly like oh yeah obviously great that that totally makes sense um and then i think the thing that was always in the back of our mind, though, with people like Google and kind of all the public clouds

was like, these are very large companies. They have very wide product portfolios. And these data warehouses are really strategically important products to them. And so what are they willing to do in order to drive more success of those data warehouses? And one of the things is obviously tightly integrate with a BI tool, but we always had in the back of our mind, are they going to try to build out their own connectors or directly compete with us? And there's been bits and pieces of that over time and they've built out more, but that was definitely one of the things that i guess maybe slightly going against how i started this answer like that was always in the back of our head just what uh where are they going and what do we do if we're going head to head for

them rather than you know just a pure partnership and complementary play so that definitely influenced our our thoughts about uh the the total long-term opportunity for us and our defensibility and stuff like that. So Stuart, so you and I were really interested in this at the time. So we'd seen a similar thing happen back in the past in the Oracle world with Thomas Kurian leading things, where the suite of Oracle BI products had been built out and extended into things like um applications and

vertical apps and so on um but but look it was look it was an interesting choice as well because it's it's quirky um it's some of some of the features about it are different but you know we we definitely we definitely saw it as being interesting what was your thoughts at the time yeah i remember the announcement happened we were exhib exhibiting at Snowflake Summit. I think it was the last day of Snowflake Summit when that happened. Yeah. Wasn't that a really nice time announcement for us too? I went to lunch with Nick Amabile when that happened. I was like, I don't want to go back.

I'm sure. And it's funny. We had had the partner, I guess, event, which ran prior to Summit itself. And there were so many shout outs back and forth between Looker and stuff like at that partner Summit. And then, you know, fast forward two days and Looker's a part of Google. And then when you look at episode 69 specifically,

great episode, I thought, between you, myself, and Tristan. Unbelievable guests. Thank you. You guys are amazing. I think it's funny. I may be wrong about this because I haven't listened to it in a few years, but that's back when they were still Fishtown. And he described Fishtown as being a smaller version of Red Pill Analytics. So I just think that's so funny to think about today, right? But, you know,

I think that it made a lot of sense, you know, because, you know, I love Snowflake, but if you looked at the overall cloud platforms, Google was my preferred cloud platform. And anything that we built at Red Pill Analytics, we built in Google. We didn't necessarily go to AWS that as much as, as others. So it, you know, it made a lot of sense now, you know, uh, here's the naysayer coming out. Right. Um, it's still, I believe not really integrated, um, in GCP. It might be, I haven't looked in a while. Uh, so forgive me Keenan, if, if that's not the case, but I haven't been there in a year and a half. It might be, I haven't looked in a while. So forgive me, Keenan, if that's not the case, but. I haven't been there in a year and a half. So. And what I've talked about on episode

69, which kind of floored Tristan a little bit, if I remember correctly, is thinking about getting Looker priced on utilization. That was the dream, right? Was that no license, procurement or anything, just spin it up and use it. And that was the thing that was, and I said in that episode, that's the thing that's going to change everything. And I don't think we're there yet. So we might be, but I don't think we're there yet. So- No, not on that front. But, you know, you know, Anurag Gupta actually from AWS asked us that almost every single time we met him. Can you start charging on a consumption basis? We thought very long and hard about that as an independent company. I think it would have been the death knell for us. It would have been very hard. That would have been a big bet the company moved on a company that was already working. So we never did that. yeah no i i agree with you steward i mean i think i think it's a big opportunity whether looker takes advantage of it under under google or or someone does right but the consumption on that would be really interesting so actually you know working with and uh and trying to recommend looker in for a project now

is quite tricky you know we um they won't pick the phone up to us now with opportunities um because um you know unless you're a GCP partner that can sort of sell all the rest of their stack, you don't qualify as being a full Looker partner. And things like customer support is different now. It's all offshore now. It's not the same kind of department of customer love. And obviously Looker is doing well as a product

and its distribution is great. But it's a different sort of environment, a different kind of partner environment than it used to be, unfortunately. No, I mean, just to be honest, I think the one big thing that I'm kind of bummed about or unexcited about is exactly what you just mentioned, Mark. Looker, we talked about this earlier,

I do think fundamentally Looker's culture was a big aspect of our success, not just internal employee culture, but our culture that we had around our customers. And our customers really felt that change massively. I mean, you can't just get on to chat and talk to some really smart analytics engineer now and debug your problems in a second in the middle of the night or something like that, right? You don't have that same hands-on level, not just in terms of just like customer support or customer success, but like the events that we would do, the community that we developed, you know? And I think DBT has done a great job of being the steward of a community

for sure around a lot of this stuff and giving folks a really awesome home. And, you know, that's the big thing I'm bummed that didn't continue. So I think it could have had a really awesome home and you know that's the big thing i'm bummed that didn't continue um because i think it could have had a really good opportunity so yeah the interaction with looker as a company is gone right and so now it's really the interaction of looker as a product and so i think you'll see a lot more of that product strategy come true as how is it does it exist in a suite of 110 products versus being its own you know its own company standalone with its own customer operation the other big bit of news that we didn't mention on that on that on that particular episode because it hadn't actually happened yet at that point um but was um was the fundraising that deep

or fishtown analytics did initially and then it obviously the company then turned into dbt labs and now it's now obviously dbt labs has become the the big the big player really in in in the modern day stock market so so um jake how does it feel to be a relative failure uh compared to uh compared to tristan and dbt labs they've done really well yeah coming in hot i was like wait a second did my dad join this call what's going on here oh no i no. I mean, it's been amazing. And I'm super pissed off about it. And ironically, totally random. I had dinner with Tristan last night.

So you could tell on what bad terms we're on. And he was hacking out of my house afterwards. But I was unhappy. Did you poison him at dinner? Oh, yeah. I made him pay is his um no it's um it's really interesting just because like the the original original origins of this was like we did these benchmark reports at rj metrics and it was like okay we've got hundreds of e-commerce companies on our platform

why don't we collect all their data together and then show some stuff about like e-commerce and aggregate and you know it was a set of tools that was like built internally primarily by our marketing team and tristan was our vp of marketing um to you know do that stuff better and like i mentioned before when when you know first tristan and then you know drew and and connor uh told us they were they were going to do this. We thought, oh, that's cool, but obviously had no idea how big it would become. And from my perspective, I think it's all upside for us. It's kind of like when the college I go to moves up in the rankings. It's like, wow, I didn't do anything for that, but I sort of seem cooler by association. So I'm happy about it. And I think one of my just

big takeaways of, I spent probably like, it's a little over a decade, maybe 12 years working on two data startups. And now I'm in a fairly different space in contracts and legal stuff. And I think just the markets are just much bigger than I expected, which is great, which just means the companies have more runway to grow and things that I thought were relatively narrow, turns out there's just a ton, a ton of people who need it. And like when Fishtown was just this consultancy, um, again, that, that made total sense to me because we were talking to people at Stitch and RJ who had these specific problems. Uh, and then it's really credit to the, what was then Fishtown and then DBT team that they had the vision of saying, okay, there is this product here that,

that is, that is a bigger opportunity bigger opportunity. Because that's not an easy transition to make. And I'd be interested to know, Stuart, if you ever thought about doing this with your firm, to make that jump from a consulting to a product company is like a tight rope walk. And obviously, they've done it well. So I think they deserve... Outside of just having a good idea, it's a lot of like really, really smart. Yeah, I mean, we tried several times and nothing we built really took off. So, you know, shout out to Fishtown now, DBT Labs for executing on that. And I remember when they first started going in that direction and sending a lot of their consulting requests to us, which was obviously great for us at the time.

This is the hardest part of the market. And it's why I'm now at a DBT competitor. And that is because this is where the rubber hits the road. It's the hardest part of the market. It's the hardest part still. Data. It's the hardest part still. Like data ingestion, the problem that Stitch and Fivetran solved was the hardest problem. They made it basically click and enable. That was huge. I can't tell you on-prem how much of a problem that was. But once that was solved, this is now the biggest

problem. Not the problem, but the biggest part of either a migration or a new implementation. So understanding that at the time they did, obviously a lot of foresight there and some smart guys figured that out. And the reason I'm in the market and the reason I'm at Coalesce now is I still think this is the biggest part of the data landscape. 90% of an analytics project, rough math there, but stay with me. 90% of the work is in that layer. Now, something like LookML takes a little bit of the, you know, hardness away. But at the same time, this is still where the meat of the project is. And this is the part of the space that's probably going to evolve

more in the future. And I think looking at interesting ways to solve this problem, I think there's going to be a lot of AI and ML approaches in the future. There's so much that a machine can figure out about your data that maybe you don't understand right away. I think we're going to, this is where the market, this is where the future of the market's real. And frankly, that's why I'm at Coalesce. Stuart, it's probably a good opportunity now to talk about what you're doing now at Coalesce. Stuart, it's probably a good opportunity now to talk about what you're doing now at Coalesce. So just maybe tell us about what the product is that you're working with and working on,

sorry, and how it kind of, I suppose, takes some of the ideas that we've been talking about and extends them a bit further, really. Yeah, Coalesce is, you know, we've spent a lot of time talking about DBT and I've gritted my teeth. Now, you know, I was a believer in their product and love how they shape the market. And now we just simply want to, you know, take them on and see if we can do it better. But Coalesce is a, you know, data transformation tool. We solve that column in the modern data stack. We do it a little bit different. We try

to meet enterprises where they are. So we have a graphical user interface. We have built-in column lineage. And underneath the covers, it's a lot of YAML and a lot of Jinja. But at the end of the day, we can abstract that away from those developers that aren't interested in that. And we can surface it for those developers that are. And, you know, I think the fact that, you know, graphical user interfaces get a, get a bad rap because of how poorly they were implemented on prim, but they're not the reason why you can't be flexible. So, you know,

that's what, that's what Colas does. And we're growing at a good rate and really excited to be back. And this is where my heart has always been. I'm a data engineer at heart. I've done a lot of analytics in Looker. I've done a lot of database design in Oracle and Snowflake. But at the end of the day, this is the part of the modern data stack that interests me the most and really excited to be running all of, you know, customer success and customer experience for Coalesce. Let's move on now to the fifth episode of the podcast we're going to talk about, which was episode number 89, Firebolt and the History of Cloud Data Warehousing

with special guest Eldad Farkash. So Keenan, obviously you're involved with Firebolt and the History of Cloud Data Warehousing with special guest Eldad Farkash. So Keenan, obviously you're involved with Firebolt. You're one of the founding team there, or certainly one of the team in this US. Tell us a bit about what Firebolt is and how you got involved with that. Yeah. I mean, I would say, I know maybe I'll take a different paradigm on what I think is the hardest and most complex part of the data ecosystem. But after being at Looker for eight, nine years and spending about a year helping the Firebolt founders continue to expand and grow their company, databases are the database is got to be one of the most insane engineering challenges that you can do in terms of what you have to think about

in terms of the almost infinite permutations of how someone wants to use a cloud data warehouse these days. Right. So for Firebolt, you know, I think they're they're massively fortunate to kind of have

just be the second generation, right, of a cloud data warehouse and look at all the different new technologies that have come about about building distributed systems in the cloud. And being able to make, you know, design choices that, you know, might, you know, will definitely help some on the bigger scale and lower latency side of things. But the myriad of use cases and permutations of stuff people want to do with a cloud data warehouse now is insane, right? And to catch all of those and to build a system to support all of that with

your same value proposition is an insanely challenging and really cool engineering challenge as well. So yeah, I mean, I went and helped them as one of the first investors in Firebolt, along with a lot of modern data stack companies that I'm excited to talk about towards the end here. But yeah, just wanted to help them build and grow and saw a lot of the internals of how to build this stuff. And so we had a fun year and a half with those guys and still continue to help them grow. So back over to Stuart, one of the things, another kind of trend that's been happening within the industry, a big sort of area of interest has been metrics layers and semantic models. Really the central role they've been playing in the ecosystem,

I suppose the balkanization of the industry a little bit around certain, around say with dbt labs and around say Google Cloud with Looker. What's your thoughts on semantic layer and metrics layers and so on at the moment yeah it's uh you know customers are going to have to choose whether they think this should be part of their you know data transformation or data engineering solution or whether it should be part of their analytics solution right for me it's always made sense as part of the analytics product. And that some of that comes from my background with Oracle Analytics, which, you know, had one of the most

sophisticated at the time semantic layers there was, it made a lot of sense to me to put that in the hands of the people that were going to consume it, not necessarily, you know, it's not always the case, right? There's a good portion of looker stuff that was built by, we'll say, practitioners and not users. But we saw lots and lots of Looker implementations where the users were doing a lot of that. So it's a difficult handoff. It's a difficult handoff between the physical structure and the eventual consumption of that physical structure. I think this needs to be solved. There's various ways to solve it, but it certainly is something that brings a lot of value to the end user. I think data catalogs play a huge part in this also,

because if what you're looking for is to understand your data. It doesn't necessarily have to be built. It can be inferred. So it'll be interesting to see where data engineering, analytics, and data catalogs all sort of butt up against each other. Who comes out in the end owning the semantic layer? What I expect to happen is just like all of these tools, you'll have a choice. I don't think there's going to be a clear winner and where it should be. I think that each organization is going to have to make a choice about which part of the stack they want to own semantics. And I don't think there's necessarily a terrible choice there.

And it's probably a one size does not fit all type of scenario. So another interesting trend and an interesting product in the market that's come along recently is Omni, of course. So we had Colin Zima come on to the show a few episodes ago. Colin was obviously at Looker before um and was uh talking about how omni's approach to analytics is actually almost the opposite of what we've been doing with with kind of planned governed semantic models where the the semantic model actually kind of you know i suppose it emerges as you start to do reporting it's not a mandatory part of what things are how you do things and it's almost going

back to actually to the approach that uh garner was talking about where the data model and the semantic model are actually optional and um really emerge out of the um the reports that you're building so you know what do you think about that i mean to keenan maybe first of all yeah i mean i think obviously you know colin and jamie were at Looker for a very long period of time, including being some of our very first customers. Do you feel betrayed by that? No, I feel like if you're betrayed by evolution, then, you know, you probably didn't evolve yourself. No, I mean, I think it's a really interesting idea, right? I mean, they're still there, you know, they're going to be massive, you know, contributors to, you know, how dbt and coalesce in the in the third party semantic ecosystem grows, as well as building their own inside of the analytics tool. And I completely agree with Stuart, this is an

extension of what we've philosophically believed at Looker is there should be choice, right? And being the best of breed things, you know, other things are going to come out of that. So the modern, you know, fast forward three years, right? Like the amount of products in the modern data stack have just really, I'm excited because it's just expanded. Massively grown, right? And it's kind of fragmented and some things won't exist, but people have really tried to innovate

on having best of breed things for all these new workflows, right? So going back to Omni, like I think it's a cool new approach, right? It helps data teams get going quickly. That was a big problem with Looker, right? But also Looker's power in semantic data modeling, they still are doing that, right? And so whether, you know, it's inside of their tool,

whether it's dbt, coalesce, or, you know, maybe, you know, Malloy starts getting traction, and maybe others started using Malloy,oy whatever i think that's a really nice approach to be doing this uh where the data teams can do it get a lot more value out of the data a lot quicker so i think there's a lot of things that they're doing at omni to to speed that up where we were a little bit front front loaded at looker and i think that's that's definitely not a secret. Jake, in some respects, you know, you voted with your feet when you founded Common Paper and that you moved out of the analytics industry and into something, a new area. So what was your take on where the industry is going and trends and so on? Yeah, sure. And I think it's in a lot of ways, you know, maybe you uh weight my opinions on

this significantly lower than uh than keenan or stewart just because i'm uh in addition whatever faults i normally have i'm like a couple years out of date now um but but i would say that like like i said before like it turns out these markets are at least much you know just much bigger than than i think we realized and that that point on best of breed like this is this was why the evolution from rj metrics to stitch made sense from like the the old tightly coupled stack into the the composable things is because you know as a company grows some people need bi dashboards some people need a python notebook some people need something totally different they need excel and it's better to to have the right tool for the job as long as there are you know good consistent interfaces that all those tools can plug into because the the the

downside that the risk is um is around you know everybody has a different version of the truth which is you know what we're all trying to, or whatever we're trying to solve with the semantic layer and things like that. And the parallel actually, with what I'm doing now, is that, in a lot of ways, the thing that got me more interested is, okay, these concepts around, like, structured data, the power of cloud data warehouses, the power of APIs, what can we do with this? The thing that always struck me is at Stitch for our own use, we had this great data warehouse and we could get an answer to almost any question we wanted to ask, maybe in 30 seconds, maybe in 10 seconds. And then there was one part of our business that was like,

oh man, we need a question about our contracts and that's going to take a week. And so the first post on the common paper blog is that contracts should actually be more like APIs than Word docs. And a lot of that is based on the fact of just like, I saw how good this can all be if you have the right data model, the right interfaces, the right tools. And that's, so a lot of what I'm doing now is just trying to bring that sort of framework and model just to a different industry

where I think it's painfully needed. I'd like to finish off then with Stuart's views on the modern data stack and the ecosystem at the moment. Where do you think it's going, really? And will open source, for example, have a big part to play in things going forward? Yeah, open source definitely. You know, if you look at a lot of the technologies that we've described in this episode, they've had an open source aspect to them. Both Stitch and DBT did. And there's a lot of things in the space that,

you know, the jury's out on that model. You know, at Coalesce, we don't have an open source aspect because we're trying to sell a great product to needy customers. And so, not needy, not needy customers in need customers in need. We'll say, I think, you know, every time I do a podcast, I say this will finally be the year of data quality. And I've been wrong so many times. That's why I, yeah, I helped found a data quality data quality company because I believe that. And so one of the things is as we move more toward data products, we'll say, whether you internally look at your data platform and what you're delivering as a product, I guarantee your customers, whether they're external or internal, do. And as we look to build products and products either work or they don't. And I think data quality is an area that we used to be able to ignore, but a product either work or they don't. And I think data quality is

an area that we used to be able to ignore, but a product either works or it doesn't. An ML model that predicts something either predicts it or doesn't predict it. So I think finally data quality is going to be a big part moving forward. And I think wrapped around that is governance. And I think semantic layer plays a part in that. I think data catalogs play a part in that. So it's going to be very interesting. We solve the really difficult stuff. Databases like Snowflake, BigQuery, and Firebolt do so much heavy lifting for us that we used to have to do ourselves. So we've solved that. Now, the data engineering side of it, there's obviously going to be innovation there. And can we reduce the friction for developers in that space and then also the the front the

final consumers which at the end of the day are the ones that drive all of this and can we make their connection to data an easier one uh you know something that they can just you know sorry for the analogy but swipe right to get right so right? So can they easily consume this and not have the sort of scenario that Jake described where, oh, suddenly I see it's going to be a week before I can get this answer. So if I look at it sort of holistically, I think we have to continue to reduce the friction in these different layers. We've done it in certain ones. We've made progress in all of them. But at the end of the day, we need, we just need to, you know, our space needs to reduce friction on a daily basis. I think the swipe right to get would be the world's most disappointing data gap.

Unless you are pre-matched, unless you're pre-matched with the data tool, then you're going to be happy. So thanks to all our guests for coming on the show. Particularly Jake, relatively short notice. So thank you there. So how does anybody find out about your latest activities? Coalesce.io. And I'll provide you some stuff for the show notes, Mark.

We've recently on partner connect with Snowflake. So you can literally in a few clicks have Coalesce up and running give it a try let us know what you think uh and uh and that's it that's how you get there and caden how do people hear about firebolt uh i even read about go to firebolts yeah there's plenty of stuff to read about go for the search yeah exactly lots uh lots lots out there and jake common paper uh yeah go to commonpaper.com follow me on uh tiktok you know all the places that last part was a joke but um yes no it's not no it's not all right i do have two videos i posted Thank you. Terima kasih telah menonton! Ketika kita mengambil alat-alat, kita bisa mengambil alat-alat yang diberikan kembali ke kota. Thank you.